In the present study, we found decreased total and HMW adiponectin levels in Latino compared to non-Latino white patients with CVD risk. Lower adiponectin levels in the Latino group were independent of BMI and other factors known to affect adiponectin, and seemed to account for the increased insulin resistance observed in this group compared to non-Latino whites. Ethnic group differences in HMW adiponectin paralleled those observed in total adiponectin, and there were no differences in HMW/total adiponectin ratios in the two ethnic groups. We identified correlates of total and HMW adiponectin levels in the two ethnic groups, confirming associations of adiponectin levels with age, gender, adiposity, and thiazolidinedione use previously reported in majority populations. Though there was no association between BMI and adiponectin on univariate analyses, the expected negative association between these two variables was observed on multivariate analyses, suggesting a blunting of the relationship due to the effect of other modulators not corrected for on univariate analysis. None of the factors found to be associated with adiponectin accounted for ethnic-group differences in total or HMW adiponectin. Finally, we did not observe the expected negative association between waist circumference and adiponectin in Latino men, suggesting differential regulation of adiponectin in this group.
Our findings of decreased adiponectin in Latinos are similar to those from studies of adiponectin in other minority ethnic/racial groups, where adiponectin is lower in minority groups studied compared to white populations [5
]. Ours is one of the first studies to explore the relationship between adiponectin and several potential regulators of adiponectin in Latinos. Though obesity is known to be associated with decreased adiponectin, differences in adiposity did not explain decreased adiponectin in the Latino group in our study. Furthermore, none of the other potential regulators tested accounted for the observed ethnic differences in adiponectin. Therefore, we conclude that decreased adiponectin in Latinos is determined by genetic or environmental factors we were not able to measure.
Studies on the contribution of adiponectin to differences in insulin sensitivity in Latinos versus non-Latino whites are still lacking. Hanley et al
reported lower adiponectin levels in African Americans compared to Latinos [31
] but comparisons between Latinos and non-Latino whites were not made in that study. Similar to our observations in Latinos versus non-Latino whites, the ethnic difference in adiponectin levels between the African American and Latino groups was no longer significant when insulin sensitivity was taken into account, demonstrating a very close association between these two variables. Contrary to this report, Latasha et al reported lower adiponectin levels in African American women compared to white women with similar insulin sensitivities [32
]. Ethnic and racial minority groups (including Latinos) participating in the Diabetes Prevention Program (DPP) study were reported to have lower baseline adiponectin levels than non-Latino white participants [5
]. Though adiponectin levels were negatively associated with HOMA-IR in the DPP population as a whole, the association of these two variables was not measured within each racial/ethnic group, and the contribution of decreased adiponectin levels to racial/ethnic differences in insulin sensitivity was not assessed. Our analyses demonstrate that the variability in adiponectin levels in Latinos versus non-Latino whites accounts for the differences in insulin resistance observed in the two groups.
Indo-Asian women have been found to have lower HMW/total adiponectin ratios than whites, suggesting an ethnic-specific regulation of adiponectin isoforms [33
]. Contrary to this report however, we found HMW adiponectin to be decreased to a similar extent as total adiponectin in Latinos compared to non-Latino whites. In addition, HMW/total adiponectin ratios did not differ between the two ethnic groups. This observation suggests the involvement of mechanisms regulating adiponectin production, with sparing of mechanisms involved in adiponectin multimerization. Consistent with observations in majority populations [34
], HMW adiponectin was more strongly associated with insulin sensitivity than total adiponectin in our population. However, both total and HMW adiponectin levels were associated with insulin sensitivity, and the ethnic-specific difference in insulin sensitivity was no longer apparent when either adiponectin measure was controlled for in our regression models. To our knowledge, ours is the first study to compare HMW adiponectin in Latinos and non-Latino whites and to identify determinants of HMW adiponectin in Latinos.
While we are not able to establish a causative association between adiponectin and insulin action in this cross-sectional study, multiple studies have shown that adiponectin has insulin-sensitizing actions (reviewed in [35
]). In addition, multiple prospective studies and a recent meta-analysis have demonstrated that individuals with low adiponectin are at higher risk for development of type 2 diabetes [5
]. In the DPP study [5
] baseline adiponectin was a strong independent predictor of incident diabetes in the three treatment groups (lifestyle, metformin, and placebo), with a hazard ratios of 0.61-0.79 per 3 microgram/ml higher adiponectin). Thus, our findings suggest that decreased adiponectin in Latinos plays a role in the high incidence of insulin resistance and type 2 diabetes in this population. As in other populations, the HMW form of adiponectin is a better determinant of insulin sensitivity than total adiponectin. However, the observed decrease in adiponectin in the Latino population is not isoform-specific.
We found an expected negative association between waist circumference and adiponectin in women from both ethnic groups, and in the non-Latino white men, but no correlation between these two variables in the Latino men group. Though we used waist circumference as a surrogate measure of visceral adiposity, our findings are consistent with those of Hanley et al
who reported an inverse correlation between visceral adiposity and adiponectin in African Americans but not Mexican Americans [31
]. These findings may indicate ethnic differences in adiponectin regulation. For instance, it is possible that Latino men are more sensitive to inhibitory visceral cytokines, resulting in an early plateau of suppressed adiponectin levels at a lower waist circumference. Alternatively, other (non-visceral) inhibitory factors may have a stronger influence on adiponectin levels in Latinos.
Our study has several limitations. Our population consisted of mostly older, overweight or obese adults, who had a diagnosis of hypertension and one additional risk factor for CVD. Therefore, our findings may not apply to a younger, leaner, or healthier population. Also, our Latino group was largely comprised of Mexican Americans, and thus our results may not be generalizable to other Latino groups. As this was a cross-sectional study, we cannot determine directionality in the correlations observed or fully explore the causes of low adiponectin in Latinos. Additional limitations of the study include our use of the homeostasis model, a surrogate marker of insulin sensitivity, rather than a more direct measure such as a frequently sampled IV glucose tolerance test or a hyperinsulinemic euglycemic clamp. Finally, we used waist circumference as an estimation of visceral adiposity rather than a more direct measurement such as can be obtained with CT scans.